help > HRF Weighting vs. Task/Condition Weighting?
Showing 1-6 of 6 posts
Apr 12, 2016 11:04 PM | Jeff Browndyke
HRF Weighting vs. Task/Condition Weighting?
It appears that CONN defaults to HRF weighting for GLM weighted
analyses, but what does the "Task/Condition" weighting option do at
the 1st level? Appropriate for task-based, block-design GLM
analyses?
Thanks,
Jeff
Thanks,
Jeff
Apr 13, 2016 09:04 PM | Alfonso Nieto-Castanon - Boston University
RE: HRF Weighting vs. Task/Condition Weighting?
Hi Jeff,
The task/condition weighting option defines which weights are used in the "weighted-GLM" analyses. When using, for example, correlation measures, a weighted-GLM analysis will simply estimate the connectivity associated with a given task/condition by computing a weighted correlation across all time-points, using as weights (by default) the hrf-convolved task response (i.e. a measure of how each scan/acquisition is expected to be affected by each task/condition). Different conditions will simply use different weights (based on each condition onsets/durations). In the first-level analysis window you can choose between three types of weights, either "hrf weighting" (the one described above, where the blocks defined by each task/condition are convolved with the canonical hemodynamic response function to define the corresponding condition-specific weights), "hanning weighting" (within each task/condition block a hanning window is used to emphasize those scans/acquisitions at the center of each block and de-emphasize the ones at the borders), or "none" (the box-car 1/0 responses defined from each task/condition onsets/durations are used instead directly as weights, without additional convolution or within-block weighting). Typically this sort of weighted-GLM analyses are perfectly fine for task-related analyses of block designs, while gPPI are typically preferred for task-related analyses of event-related designs.
Hope this helps
Alfonso
Originally posted by Jeff Browndyke:
The task/condition weighting option defines which weights are used in the "weighted-GLM" analyses. When using, for example, correlation measures, a weighted-GLM analysis will simply estimate the connectivity associated with a given task/condition by computing a weighted correlation across all time-points, using as weights (by default) the hrf-convolved task response (i.e. a measure of how each scan/acquisition is expected to be affected by each task/condition). Different conditions will simply use different weights (based on each condition onsets/durations). In the first-level analysis window you can choose between three types of weights, either "hrf weighting" (the one described above, where the blocks defined by each task/condition are convolved with the canonical hemodynamic response function to define the corresponding condition-specific weights), "hanning weighting" (within each task/condition block a hanning window is used to emphasize those scans/acquisitions at the center of each block and de-emphasize the ones at the borders), or "none" (the box-car 1/0 responses defined from each task/condition onsets/durations are used instead directly as weights, without additional convolution or within-block weighting). Typically this sort of weighted-GLM analyses are perfectly fine for task-related analyses of block designs, while gPPI are typically preferred for task-related analyses of event-related designs.
Hope this helps
Alfonso
Originally posted by Jeff Browndyke:
It appears that CONN defaults to HRF weighting
for GLM weighted analyses, but what does the "Task/Condition"
weighting option do at the 1st level? Appropriate for
task-based, block-design GLM analyses?
Thanks,
Jeff
Thanks,
Jeff
Apr 14, 2016 02:04 PM | Jeff Browndyke
RE: HRF Weighting vs. Task/Condition Weighting?
Hi, Alfonso.
Under the ROI-to-ROI and seed-to-voxel analysis options there are actually four choices:
1.) no weighting
2.) HRF weighting
3.) Hanning weighting
4.) Task/condition factor
It was the fourth weighting choice (i.e., task/condition factor) that I was wondering about. What is this weighting choice? And, should it be used for task-based FC analyses?
Thanks for the guidance and advice.
Warm regards,
Jeff
Under the ROI-to-ROI and seed-to-voxel analysis options there are actually four choices:
1.) no weighting
2.) HRF weighting
3.) Hanning weighting
4.) Task/condition factor
It was the fourth weighting choice (i.e., task/condition factor) that I was wondering about. What is this weighting choice? And, should it be used for task-based FC analyses?
Thanks for the guidance and advice.
Warm regards,
Jeff
Apr 15, 2016 12:04 PM | Alfonso Nieto-Castanon - Boston University
RE: HRF Weighting vs. Task/Condition Weighting?
Hi Jeff,
Sorry for misinterpreting. Yes, the (4) option simply allows manually defining the weights using your own externally-defined timeseries. You may enter a timeseries of weight values separately for each condition in the Setup.Conditions "condition modulation factor" field, and then indicate that you want to use those timeseries as weights in the weighted-GLM first-level analysis procedure by selecting the 4th option (task/condition factor) in the "weights" field there. Typically for task-based FC analyses the default "hrf weighting" choice (option 2) is the recommended setting (where the weights are implicitly defined by convolving the task blocks with the hrf response). The additional 4th option there simply extends this to cases where you may prefer to use a different form of weighting function that is not covered by any of the three options above.
Hope this helps
Alfonso
Originally posted by Jeff Browndyke:
Sorry for misinterpreting. Yes, the (4) option simply allows manually defining the weights using your own externally-defined timeseries. You may enter a timeseries of weight values separately for each condition in the Setup.Conditions "condition modulation factor" field, and then indicate that you want to use those timeseries as weights in the weighted-GLM first-level analysis procedure by selecting the 4th option (task/condition factor) in the "weights" field there. Typically for task-based FC analyses the default "hrf weighting" choice (option 2) is the recommended setting (where the weights are implicitly defined by convolving the task blocks with the hrf response). The additional 4th option there simply extends this to cases where you may prefer to use a different form of weighting function that is not covered by any of the three options above.
Hope this helps
Alfonso
Originally posted by Jeff Browndyke:
Hi, Alfonso.
Under the ROI-to-ROI and seed-to-voxel analysis options there are actually four choices:
1.) no weighting
2.) HRF weighting
3.) Hanning weighting
4.) Task/condition factor
It was the fourth weighting choice (i.e., task/condition factor) that I was wondering about. What is this weighting choice? And, should it be used for task-based FC analyses?
Thanks for the guidance and advice.
Warm regards,
Jeff
Under the ROI-to-ROI and seed-to-voxel analysis options there are actually four choices:
1.) no weighting
2.) HRF weighting
3.) Hanning weighting
4.) Task/condition factor
It was the fourth weighting choice (i.e., task/condition factor) that I was wondering about. What is this weighting choice? And, should it be used for task-based FC analyses?
Thanks for the guidance and advice.
Warm regards,
Jeff
Apr 15, 2016 04:04 PM | Jeff Browndyke
RE: HRF Weighting vs. Task/Condition Weighting?
Thanks, Alfonso!
Warm regards,
Jeff
Warm regards,
Jeff
Feb 14, 2018 12:02 AM | Benson Stevens
RE: HRF Weighting vs. Task/Condition Weighting?
Hi There,
I have a question concerning the choice of "hrf-convolved task response or "none." I am doing an audio loclaizer with sparse sampling. I chose the sparse option in the basic setup page. I know that makes the conditions not convolved with the hrf. However, will choosing the "hrf-convolved task response" in the 1st level stats part then convolve them, or will it used the unconvolved ones created from the sparse sampling option? I kindly thank you in advance.
Many Blessings,
Benson
I have a question concerning the choice of "hrf-convolved task response or "none." I am doing an audio loclaizer with sparse sampling. I chose the sparse option in the basic setup page. I know that makes the conditions not convolved with the hrf. However, will choosing the "hrf-convolved task response" in the 1st level stats part then convolve them, or will it used the unconvolved ones created from the sparse sampling option? I kindly thank you in advance.
Many Blessings,
Benson